Answers are grounded in this paper and cite their sources.
Background Uncommon EGFR mutations account for a substantial minority of EGFR-mutant non-small cell lung cancer (NSCLC). Whether third-generation EGFR tyrosine kinase inhibitors (3G TKIs) outperform older options in these genotypes remains unclear, and randomized comparisons are lacking. Objective: To compare 3G TKIs with first-generation (1G) or second-generation (2G) TKIs or chemotherapy within genotype-defined subsets of uncommon mutations. Methods PubMed, Embase, Web of Science, and the Cochrane Library were searched for comparative studies. Outcomes were pooled separately for non-exon 20 insertion uncommon mutations, P-loop and αC-helix compressing (PACC) mutations, exon 20 insertions, and rare exon 19 variants using random-effects models. Risk of bias and certainty were rated with ROBINS-I, the Newcastle-Ottawa Scale, and GRADE, and leave-one-out and specification analyses tested robustness. Results Twelve non-randomized studies were included, ten with serious risk of bias. In non-exon 20 insertion uncommon mutations, the pooled objective-response comparison was inconclusive (risk ratio [RR], 0.81; 95% CI, 0.59–1.13). When reverse-counted events from the largest, Tier one inverse-probability-weighted cohort were added only as a sensitivity analysis, the objective-response risk ratio was 0.76 (0.62–0.94) and does not replace the primary pool. A disease-control RR of 1.09 (1.01–1.18) did not persist in sensitivity analyses and should not be read as an efficacy gain. Response in PACC and exon 20 insertion subsets was inconclusive. In two small 19delins cohorts, progression-free survival appeared to favor 1G TKIs (hazard ratio, 2.35; 1.52–3.63). Versus 2G therapy, 3G TKIs were associated with a lower risk of grade 3 or higher diarrhea (RR, 0.25; 0.08–0.76). Conclusion These data suggest that third-generation TKIs may not confer a class-wide efficacy advantage in uncommon EGFR-mutant NSCLC. Treatment selection remains inconclusive and might be better informed by genotype and toxicity profile, although the very low certainty of evidence calls for prospective genotype-stratified trials.
No takes yet. Share an insight, caveat, or question.
Biaoxue et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: